Bug 146475

Summary:	e100 crashing
Product:	[openSUSE] SUSE Linux 10.1	Reporter:	Terje Bergström <terje.bergstrom>
Component:	Kernel	Assignee:	Karsten Keil <karsten.keil>
Status:	RESOLVED INVALID	QA Contact:	E-mail List <qa-bugs>
Severity:	Normal
Priority:	P5 - None
Version:	Beta 2
Target Milestone:	---
Hardware:	x86
OS:	SUSE Other
Whiteboard:
Found By:	Other	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	Hardware info taken with SUSE 10.0 boot.msg from a working 10.0 install boot.msg from a 10.0 updated to 10.1 Beta 2. e100 crashes later. boot.msg taken from a fresh install of 10.1 Beta 2

Description Terje Bergström 2006-01-28 19:05:57 UTC

After updating from SUSE 10.0 to 10.1 Beta 2, the network card stopped working. The network card is assigned irq 11. I got the following message from kernel:

Jan 28 13:02:36 terje klogd: irq 11: nobody cared (try booting with the "irqpoll
" option)
Jan 28 13:02:36 terje klogd:  [<c0134d47>] __report_bad_irq+0x2b/0x69
Jan 28 13:02:36 terje klogd:  [<c0134f04>] note_interrupt+0x17f/0x1ab
Jan 28 13:02:36 terje klogd:  [<c013486a>] handle_IRQ_event+0x23/0x4c
Jan 28 13:02:36 terje klogd:  [<c013492d>] __do_IRQ+0x9a/0xcd
Jan 28 13:02:36 terje klogd:  [<c0104f1c>] do_IRQ+0x46/0x56
Jan 28 13:02:36 terje klogd:  [<c0103afa>] common_interrupt+0x1a/0x20
Jan 28 13:02:36 terje klogd:  [<c028007b>] inet6_hash_connect+0x1c/0x3cd
Jan 28 13:02:36 terje klogd:  [<c011c1b9>] __do_softirq+0x2c/0x7f
Jan 28 13:02:36 terje klogd:  [<c011c22e>] do_softirq+0x22/0x26
Jan 28 13:02:36 terje klogd:  [<c0104f21>] do_IRQ+0x4b/0x56
Jan 28 13:02:36 terje klogd:  [<c0103afa>] common_interrupt+0x1a/0x20
Jan 28 13:02:36 terje klogd: handlers:
Jan 28 13:02:36 terje klogd: [<e0c06548>] (usb_hcd_irq+0x0/0x4f [usbcore])
Jan 28 13:02:36 terje klogd: [<e0d8180d>] (e100_intr+0x0/0xd0 [e100])
Jan 28 13:02:36 terje klogd: Disabling IRQ #11

When I installed 10.1 Beta 2 as a fresh install to a separate partition, the network card worked and there was no panic.

Comment 1 Karsten Keil 2006-01-29 11:43:25 UTC

Hmm, then the reason maybe different boot options or some different module
settings/usage.
In the working configuration does it also use IRQ 11 and the e100 driver ?
Please also attach a hwinfo for your HW.

Comment 2 Terje Bergström 2006-01-29 11:57:11 UTC

Created attachment 65549 [details]
Hardware info taken with SUSE 10.0

Hardware info taken with working 10.0 install

Comment 3 Terje Bergström 2006-01-29 11:58:22 UTC

Created attachment 65550 [details]
boot.msg from a working 10.0 install

Comment 4 Terje Bergström 2006-01-29 11:59:34 UTC

Created attachment 65551 [details]
boot.msg from a 10.0 updated to 10.1 Beta 2. e100 crashes later.

Comment 5 Terje Bergström 2006-01-29 12:00:07 UTC

Created attachment 65553 [details]
boot.msg taken from a fresh install of 10.1 Beta 2

Comment 6 Terje Bergström 2006-01-29 12:04:46 UTC

I attached three boot.msg's. In each IRQ 11 was assigned to e100. The default boot options are used, and they're the same in each case. The difference was that in 10.1 fresh install a resume= option is added. No module options are provided for e100.

Comment 7 Karsten Keil 2006-01-29 12:15:37 UTC

It seems that the fresh install do not use e100 but eepro100, can you try to load eepro100 ?

Comment 8 Terje Bergström 2006-01-29 18:02:20 UTC

I have already downgraded the non-working 10.1 upgrade. This line:

<6>e100: eth0: e100_probe: addr 0xfe8ff000, irq 11, MAC addr 00:03:0D:03:7A:B6

makes me think that it's actually e100 that took over the card, and not eepro100.

Comment 9 Olaf Kirch 2006-01-30 07:31:39 UTC

The message doesn't signal a crash, but says an interrupt arrived and
was handled neither by e100 nor by the USB stack. Please compare
your modules.conf files and attach hwinfo output for both configurations.

It may not be the e100 driver that's broken - could also be USB. Note
the many USB errors in the non-working case.

Try disabling ehci_hcd (by moving the module out of /lib/modules,
or by blacklisting it in modules.conf), reboot.

Then reenable ehci_hcd, disable, e100, reboot.

This should tell us more.

Comment 10 Greg Kroah-Hartman 2006-01-30 17:34:14 UTC

There's a fix for the ehci bios issue that could cause this irq issue (but
it should show up when the ehci driver is being loaded).  It's in the kernel
of the day, if you could try that.

Comment 11 Terje Bergström 2006-01-31 19:35:33 UTC

I removed uhci_hcd and ehci_hcd, rebooted. I got the same stray irq 11 message, except that the usb_hcd_irq line was missing. I put the modules back, and removed e100. Still the same message, except of course usb_hcd_irq line was back, but e100 line wasn't there.

Comment 12 Gerald Pfeifer 2006-04-20 22:00:11 UTC

It seems this got stuck in NEEDINFO. :-(

Comment 13 Terje Bergström 2006-04-23 08:12:14 UTC

I don't have this system installed anymore (replaced with a fresh install of OpenSUSE), so I cannot provide any more logs/info.

Comment 14 Greg Kroah-Hartman 2006-04-25 05:07:11 UTC

Ok, will close this bug then, thanks for letting us know.